group work, IQ, and underperformance

re: group work lowers IQ

I've skimmed the article (free online).

Assuming the findings are confirmed in other studies (I suspect they will be), this is bad news.

Set-up

• Subjects had the same IQ: 126.
• They were put in small groups of 5 and introduced to each other.
• They took an IQ test with no feedback as to how they did.
• Then they took a second computer-administered IQ test.
• After each question, they were told whether they got the answer right or wrong.
• At the same time, they were also given their rank inside the group (rank determined by each person's # of correct answers).
• They were also given their rank vis a vis 1 particular member of the group, chosen "pseudorandomly."
• 2 people had brain scans during the test-with-feedback condition.

Results

• Everyone did worse in the beginning. Across the board. Everyone. Everyone did worse than his/her measured ability.
• As the test went on, some people recovered. Their performance went back up to the level predicted by their IQ scores. 
• The others never improved. They started low, and they stayed low. 
• Females were more likely to start low and stay low than males. 
• For high performers, brain scans showed activity in the amygdala (likely fear), decreasing over time. (That is, they were likely feeling less fear, perhaps growing more confident.)
• For high performers, activity in the lateral PFC increases over time. Activity in lateral PFC is associated with IQ tasks, with working memory tasks, and with increased task difficulty. 

So. All told, membership in a small group produced a net decrease in number of correct answers compared to what you would predict for the same 5 people working alone.

Everyone's performance dropped at the outset.

Some people recovered, others didn't.

The people who recovered simply went back up to where they had been going in, before experimenters assigned them to a small group.

Questions

• How small is small? Would group of 20 students in whole class instruction show the same pattern? 25 students? 30? 
• If so ... yikes.
• In the wake of this study, mixed-ability groups strike me as an even worse idea than I've thought in the past. Lower ability children in a mixed-ability group are going to be getting constant negative feedback about their status vis a vis the higher ability children. On the other hand, the study did not include a condition that manipulated feedback in this manner. That would be interesting.
• Assuming this study picked up on a personality difference (which we don't know, of course), what would happen if you grouped the 'nervous' kids together, putting the 'confident' kids in their own group? Would two groups still separate out in this way?
• Would same-sex groups change the results?
• What does this tell us about grades and grading? 

The blue bar represents the low performers, the red bar high performers. All have the same measured IQ (126), and at the beginning of the study all are performing well below the level their IQ would predict. The higher performers then recover, and their performance increases to a "126" level. The the low performers do not recover, and their performance remains suppressed.

Implicit signals in small group settings and their impact on the expression of cognitive capacity and associated brain responses